ABSTRACT
The integration of four artificial intelligence tools that combine the ID3 algorithm and the case-based reasoning method of the nearest neighbor are proposed to solve the problem of characterizing the flashover on high-voltage insulators. The first tool uses data mining to build a classification or decision tree from historic data, the second, generates production rules, the third, operates the decision tree as an expert system, and the last, makes tests with known cases to evaluate classification accuracy. These tools are applied to the high-voltage insulators flashover problem, and the results are compared against other machine learning tools: C4.5, FOIL, CN2 and OC1.
